On Mon, Jan 30, 2012 at 08:30:37PM +0100, René Hummen wrote:
>
> On 30.01.2012, at 12:41, Diego Biurrun wrote:
>
> > On Sat, Jan 28, 2012 at 07:09:32PM +0100, René Hummen wrote:
> >> On 25.01.2012, at 16:14, Fahad Aizaz wrote:
> >>> Fahad Aizaz has proposed merging lp:~fahad-aizaz/hipl/hip-cert-conf into lp:hipl.
> >>>
> >>> --- hipd/init.c 2011-12-13 13:50:53 +0000
> >>> +++ hipd/init.c 2012-01-25 15:13:34 +0000
> >>> @@ -465,81 +435,6 @@
> >>>
> >>> -/* Needed if the configuration file for certs did not exist */
> >>> -#define HIP_CERT_INIT_DAYS 10
> >>
> >> ... and once we are on it, we can also use this define instead of
> >> the parameter in the configuration file. This change would make
> >> hip_cert.conf obsolete.
> >
> > How can build-time configuration make run-time configuration obsolete?
>
> The generated configuration file is only used by the test application
> certteststub. My thought: Is it really necessary to offer run-time
> configuration support by means of a config file for a test
> application? I especially ask this question seeing that the current
> certificate code neither allows signaling of certificates between two
> instances of the hipd nor verification of certificates by the hipfw.
> In my opinion, having a config file for this specific test stub is
> overkill and the removal of the code responsible for reading and
> parsing the config file would enable us to further shrink the HIPL
> codebase.
OK, I understand what you mean now and agree in full - let's remove this
config file and its reading code. Note that there will still be three
other ways of dealing with config files left afterwards...
On Mon, Jan 30, 2012 at 08:30:37PM +0100, René Hummen wrote:
>
> On 30.01.2012, at 12:41, Diego Biurrun wrote:
>
> > On Sat, Jan 28, 2012 at 07:09:32PM +0100, René Hummen wrote:
> >> On 25.01.2012, at 16:14, Fahad Aizaz wrote:
> >>> Fahad Aizaz has proposed merging lp:~fahad-aizaz/hipl/hip-cert-conf into lp:hipl.
> >>>
> >>> --- hipd/init.c 2011-12-13 13:50:53 +0000
> >>> +++ hipd/init.c 2012-01-25 15:13:34 +0000
> >>> @@ -465,81 +435,6 @@
> >>>
> >>> -/* Needed if the configuration file for certs did not exist */
> >>> -#define HIP_CERT_INIT_DAYS 10
> >>
> >> ... and once we are on it, we can also use this define instead of
> >> the parameter in the configuration file. This change would make
> >> hip_cert.conf obsolete.
> >
> > How can build-time configuration make run-time configuration obsolete?
>
> The generated configuration file is only used by the test application
> certteststub. My thought: Is it really necessary to offer run-time
> configuration support by means of a config file for a test
> application? I especially ask this question seeing that the current
> certificate code neither allows signaling of certificates between two
> instances of the hipd nor verification of certificates by the hipfw.
> In my opinion, having a config file for this specific test stub is
> overkill and the removal of the code responsible for reading and
> parsing the config file would enable us to further shrink the HIPL
> codebase.
OK, I understand what you mean now and agree in full - let's remove this
config file and its reading code. Note that there will still be three
other ways of dealing with config files left afterwards...
Diego